This book "asserts that the origin and spread of languages must be examined primarily through the time-tested techniques of linguistic analysis, rather than those of evolutionary biology" and "defends traditional practices in historical linguistics while remaining open to new techniques, including computational methods" and "will appeal to readers interested in world history and world geography."
Date: Mon, 21 Jun 2004 15:17:26 +0200 From: Fredrik Heinat Subject: Skeptical Linguistic Essays
AUTHOR: Paul M. Postal TITLE: Skeptical Linguistic Essays PUBLISHER: Oxford University Press YEAR: 2004
Fredrik Heinat, University of Lund.
BRIEF SUMMARY Postal's book is a collection of 14 papers divided into two parts. The six papers in the first part deal with as diverse topics as locative inversion (chapter 1), raising from subject to complement position of P (ch.2), complex DP shift and right node raising (ch.3), chromaticity, a new category distinction of English DPs (ch.4), negative polarity items (NPI) (ch.5) and the openness of Natural Languages (ch.6). Postal's purpose with the first five chapters is to show that even in such a well studied language as English, there are still too many things left unexplained to warrant the optimistic view that 'linguists are today in possession of, or even close to, adequate descriptive accounts of syntax' (p.8). The second part consists of eight chapters dealing with studies in 'junk linguistics'. The first chapter in this part deals with strong crossover effects (ch.7). The next chapter deals with the passive voice (ch.8). Chapter 9 is a paper previously published as a topic-comment paper in Natural Language and Linguistic Theory, 'Advances in Linguistic Rhetoric'. Chapter 10 gives Postal's view of the refereeing process. Chapter 11 explores the most irresponsible passage in linguistics, according to Postal. Chapters 12 and 13 are analyses of the uses of the words 'automatically', 'virtually', 'natural' and '(virtually) conceptually necessary' in the linguistic literature. The final chapter is a list of advice how to deal with junk linguistics.
COMMENTS ON THE PAPERS The first chapter, 'A Paradox in English Syntax' deals with locative inversion structures (1).
(1) Into the room walked a knight in shining armour.
The paradox according to Postal (p.17) is that there is abundant evidence for treating the preverbal PP as the subject of the sentence (Bresnan 1994), but on the other hand, there is abundant evidence that this PP is not a subject. Rather, it behaves similar to subjects of expletive clauses. Postal's analysis of these PPs is couched in the framework of arc-pair grammar (Johnson & Postal 1980). The conclusion is that these constructions have an invisible expletive and the PP has certain properties in common with ordinary subjects.
The chapter contains a wealth of data and interesting discussions. Postal's analysis is rather technical and some familiarity with arc pair grammar is a plus for the reader. Even though Postal introduces the framework, it is difficult for the uninitiated reader to grasp all the details.
The second chapter 'A Putatively Banned Type of Raising' deals with raising to complement of P (RCP) (2).
(2) Mary can depend on Bill to do something foolish.
The DP 'Bill' seems to have raised from the subject position of the embedded clause to the complement position of the preposition, hence the name RCP. Postal takes a starting point the general consensus of three major syntactic theories (government and binding, lexical functional grammar, and head-driven phrase structure grammar) that RCP does not exist in natural language. First Postal gives the (theory internal) reasons for each of the theories above, why there cannot be such a thing as RCP. In the rest of the chapter he argues, supported by data, that there in fact is RCP and that 'theoretical convergence' does not mean that whatever theories converge on can be taken for correct.
In the third chapter 'A New Raising Mystery' Postal explores the correlation between Right Node Raising and Complex DP Shift, and subject antipronominal contexts. In general both Right Node Raising and Complex DP Shift are insensitive to subject antipronominal contexts (Postal 1994). The exception, Postal argues is when there is raising to object. More specifically, the restriction is connected to raising to object of a resumptive (invisible) pronoun linked to Right Node Raising or Complex DP shift. According to Postal it is this resumptive pronoun (RP) that is raised to an antipronominal position and causes the ungrammaticality. The structure looks as (3) (Postal's (36) p.117).
Using partly new data, Postal concludes that subject to object raising exists, that there is evidence for invisible pronouns in topicalization clefts, object raising, object deletion and parasitic gaps, and that this kind of analysis requires some kind of 'look ahead', which disfavours any kind of grammar that does not allow it.
In chapter four, 'Chromaticity', Postal introduces a new category distinction for nouns. DPs can be divided into 'chromatic' and 'nonchromatic'. The terminology is based on the fact that the nonchromatic DPs have non-specific and very general meaning. According to Postal this is a closed class, containing e.g. the following DPs: 'something', 'anything', 'nothing', 'what', 'whatever'. Postal identifies seven properties that distinguish between chromatic and nonchromatic DPs. There are also selectional restrictions. Certain verbs obligatory selects for chromatic DPs, e.g. 'speak' in (4)
(4) Mary speaks some/every language/*something/*stuff.
Certain environments require a nonchromatic DP, e.g. 'be the matter with' as in (5).
(5) Nothing/something/*old age/*mould is the matter with that cheese.
There are also agreement phenomena with e.g. pronouns, relative clauses, ellipsis and subject control. As pointed out by Postal, this chapter is very informal in nature and rather than giving a detailed analysis of the phenomenon it fleshes out the generalisations a proper analysis must account for.
Chapter five, The Structure of One Type of American English Vulgar Minimizer' deals with the syntax of vulgar minimizers, like 'squat', 'shit', etc (5).
(5) a. Mary knows shit about quantum physics. b. Mary doesn't know shit about quantum physics.
They apparently behave like 'nothing' and 'anything', but, according to Postal, contrary to intuition it does not make sense to treat these minimizers as homonyms. Instead he suggests the following structure (6)(Postal's (32) p166):
(6) squat = [DP [D Zero] + [N squat]]
where the determiner is similar to the numeral zero. He also extends this analysis to the more standard form word 'nothing'.
The last chapter in the first group of essays, 'The Openness of Natural Language' presents arguments against the possibility for NLs to have proof-theoretic/generative grammars. Postal's main point is that NLs are 'open'. Using direct speech as part of the data, he claims that the openness of NL, invalidates Chomsky's (1959) claims that sentences have finite length, a finite vocabulary, and that a generative grammar is an adequate tool for describing NL.
The second part of the book is called 'Studies of Junk Linguistics'. Of the eight papers in this section, the first two are definitely the most interesting ones (from a linguistic point of view). The first, 'Junk Syntax 1: A Supposed Account of Strong Crossover Effects' (SCE) is a critique of the attempts, or rather the lack of attempts, in government and binding theory to reduce the effects of SCE to binding principle C violations. Postal claims that these effects have never been proven and gives an extensive overview of SCE and lots of data that refutes the principle C account.
The next chapter, 'Junk Syntax 2: ''There Remain a Few as Yet Unexplained Exception''' is written in a similar vein. But here it is the government and binding analysis of passives that is under scrutiny. The starting point is the transformational account of passives given in Syntactic Structures. Postal shows that there are three major problems with this account and its variant (move alpha). First, there is overgeneration (with for instance 'buy'). Second, undergeneration (e.g. double objects and pseudopassives). Third, there is no satisfactory account of the 'by'-phrase. Just as in the previous chapter the reader gets a thorough introduction and background to the topic. There are also lots of data and arguments for why the analysis in question fails. This chapter is also a harsh critique of how the transformational account has more or less been taken for granted in government and binding, despite the fact that all data have been known, and even cited in the relevant literature.
The third chapter in the second section, 'Advances in Linguistic Rhetoric' is a reprint from Natural Language and Linguistic Theory. Of the studies in junk linguistics this is the most entertaining and enjoyable paper. The next four chapters do not focus on junk linguistics in general (even though government and binding got most of the bashing in the previous chapters), but rather on one particular linguist, which gives the reader the impression that as long as you keep out of government and binding/the minimalist program, or if you don't do what Chomsky says, you are not doing junk linguistics. This is not to say that Postal's critique is unjustified, but it is tedious reading after a while. In addition these are not the kind of papers you expect when you read the blurb. (And definitely not what you want to pay 39.95 dollars for when you don't expect to get it.)
The final chapter contains guidelines for dealing with junk linguistics. They are very general and can, and should, be applied to any science. The long and the short of this list how to avoid junk linguistics is: don't do it.
GENERAL COMMENTS This book is a goldmine for advanced students in syntax interested in, and particularly starting to work on, locative inversion, raising, strong crossover effects or passives. The chapters dealing with these topics give thorough theoretical background, often from more than one theoretical perspective, and a host of data and references. The last part of book gives the reader something to think about concerning ethics in science, but is slightly tiresome after a while. Even though the criticism of Chomsky is justified the reader keeps wondering if junk linguistics is only carried out in government and binding/minimalist program or if it exists in other generative frameworks as well, not to mention non- generative ones. Some previous knowledge of syntactic theory, particularly government and binding/minimalism, is a plus in order to fully appreciate Postal's criticism. To conclude, I take Postal's advice to students to my heart; Beware, be sceptical.
REFERENCES Bresnan, J. 1994. Locative Inversion and the Architecture of Universal Grammar. Language 70:72-131.
Chomsky, N. 1959. On Certain Formal Properties of Grammars. Information and Control 2:137-167.
Johnson, D. E. & P. M. Postal. 1980. Arc Pair Grammar. Princeton, New Jersey: Princeton University Press.
Postal, P. M. 1994. Parasitic and Pseudo- Parasitic Gaps. Linguistic Inquiry 25:63-117.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER Fredrik Heinat is a Ph.D. student at the Department of English at the University of Lund. His research interests are the syntactic structure of pronouns, reflexives, and binding theory.